The RATS Software Forum

Hi

In the article ”A small sample correction for the test of cointegrating rank in the vector autoregressive model” (Johansen, Econometrica, 70/5, 2002), it is told that a RATS program is given for calculating the correction factor in Johansen – Hansen – Fachin (2002)*. In CATS2, the correction is readily available, but only in cases where the maximum lag is 4 or smaller. I have tried to download the Johansen et al. (2002) paper and to find the RATS program through Estima pages and various RATS manuals, but have not succeeded.

I wonder whether there is such a code readily available?


*A simulation study of some functionals of random walk. / Johansen, Søren; Hansen, Henrik; Fachin, Stefano.
Københavns Universitet, 2002.
Publication: Research › Working paper

We put the block on lags bigger than 4 because it takes a long time to do the calculation for big models. (We had people for whom it took an hour to do the small sample corrections). It requires computing the ergodic variances for a state space model the size of the short-run VAR (thus N*N*L states) which gets to be a very big system of equations very quickly. If you have the patience to do bigger calculations, change the bottom line shown in your cats.src file:

*** Note: Added the following global to facilitate skipping
*** the Bartlett Rank correction beyond a certain number
*** of lags--to avoid slow execute/memory limitations that
*** can arise at longer lag lengths:

compute %MaxBartlettLag = 4

Many thanks Tom!

I need the correction for small samples with up to 4 lags. Can you please provide me with the link to the RATS procedure that makes this correction. I need it desperately.

Many thanks,
Petros

Petros wrote:I need the correction for small samples with up to 4 lags. Can you please provide me with the link to the RATS procedure that makes this correction. I need it desperately.

Many thanks,
Petros

You don't have to do anything. The block is on doing the correction with more than 4.